Bioz

Bioz is an AI-powered literature discovery platform for life sciences, using machine learning to index and rank experimental findings from scientific publications to accelerate biomedical research.

Bioz is a commercial platform that applies artificial intelligence to the discovery and interpretation of scientific literature in the life sciences. Founded in 2013 and based in Los Altos, California, the company develops software tools to help researchers identify reagents, protocols, and experimental products from published studies. The platform's primary output is a searchable database of biological and chemical products, each annotated with evidence from full-text articles, structured for experimental context.

The company was founded by Karin Lachmi, who serves as CEO, and Ori Perlman, who serves as CTO. Prior to Bioz, Lachmi worked in academic research and co-founded genetics startup which informed her goal to build a tool that would reduce the friction of finding reliable experimental materials. Perlman had a background in software engineering and data analytics. The company raised about $20 million across several funding rounds, with investors including Intel Capital, which participated in a $12.5 million round in 2019.

Bioz focuses its technology on the experimental reproducibility problem. Millions of research papers contain hidden or incomplete descriptions of products, which makes it hard for scientists to reproduce results. Bioz searches the full text of publications, and it uses machine learning and other deep learning techniques to extract and analyze mentions of products, such as reagents, antibodies, kits, and equipment. It then scores those products to show their predicted likelihood to work in a given experimental context.

Platform and Technology

The Bioz platform indexes tens of millions of data points from over 25 million life science research articles. Using large language models and other AI approaches, it processes the articles to identify product names, their relationships with protocols, and the outcomes of experiments. The system integrates search tools that allow researchers to enter a query, and the AI generates results that go beyond simple keyword matching. It ranks those hits by a proprietary "Bioz Score," which reflects the predicted quality of evidence for each product used in a specific context.

The platform offers several key products, including Bioz Search, Bioz API, and Bioz Solutions. Bioz Search is the free web interface that lets users look up products and get contextualized results, including summaries of evidence extracted from papers. Bioz API allows organizations to integrate these search and scoring features into their own internal systems, such as electronic lab notebooks or laboratory information management systems. Bioz Solutions is a suite of data-driven tools made for life science vendors - they are used to help them improve their marketing and their e-discovery strategies.

Bioz also provides a Chrome extension that is an AI-powered assistant. This tool helps researchers in real time - it detects product mentions on web pages and in the browser, links them to the underlying evidence, and gives users direct access to relevant data without losing their browsing context.

AI and Technology

Bioz's approach relies on a combination of generative AI, deep learning, and neural network architectures. The founders describe the technology as akin to building a Google for life science experiments, but with the ability to understand the context of how and how to use products. Instead of simply returning all articles that mention a product, Bioz's system identifies specific things - such as a study's methods, the type of assay, species used, and experimental outcomes that are relevant.

To achieve this, Bioz develops and uses custom models that process full-text PDFs and turn them into structured JSON of experimental metadata. This includes details like paired agents, reagents, equipment, and the specific experimental conditions. The company internally refers to the system's targets as several types of 'deep machine learning models,' which are trained on publicly available articles and on proprietary data sets that are generated and labeled by a team of scientists.

Bioz has received recognition from the AI community: in 2019, the company was the winner of the Amazon Hinds AI Innovation Challenge, and in 2020 it was selected for the NVIDIA Inception program. These honors highlighted its specialized use of GPU computing and its innovative application.

Integration and Partnerships

Bioz has formed alliances with major institutions and publishers. In 2019, it announced a partnership with the Nature Publishing Group platform to provide the underlying data for a new life sciences discovery tool. The following year, the company collaborated with the Google Cloud to expand its capacities and demonstrate the scaling of its cloud-based AI workflows.

In 2021, Bioz introduced an integration with MilliporeSigma, a major life science vendor, which allowed the company to connect its Bioz Score into the vendor's own site so that customers could see the predictive performance of an item. This partnership was significant in shaping the company's business model from selling to marketplaces and vendors to use their data for targeted product recommendations.

The platform also serves academic institutions and biotechs. Its users include research groups that want to quickly identify the best tool for a new experiment and vendors that want to know how their products are being cited, to provide their marketing teams with evidence.

Funding and Growth

Bioz raised $20 million, according to public records. Its first major funding round was in 2016 as a seed round, including contributions from private investors. In 2019, the company completed a $12.5 million Series A round led by Intel Capital, which was noted as an investment in AI for the life sciences. In 2023, Bioz raised an additional undisclosed round, increasing its total to about $20 million.

The company has offices in the US and in Israel, which is where its research and development team is located. Its main headquarters is in Los Altos, California.

Challenges

One of the challenges Bioz addresses is the broad and unstructured nature of life science literature. Published studies often omit precise details, such as the exact concentration, vendor, or catalog number for the products, which makes it hard to reproduce experiments. Bioz's AI works to fill in the gaps, combining text extraction with domain knowledge to generate structured metadata. However, the quality of the extracted data remains a concern for the future, as do the careful approaches to validating the accuracy of its model outputs.

Another challenge is integrating its proprietary scores with the rapidly changing field of research. Bioz continuously learns from newly published articles, which means its scoring models can improve over time. Yet, questions remain about the extent to which a product's past performance predicts its future usefulness in an experiment, and the interpretability of its scores.

Future Directions

Bioz is positioned to be part of the broader movement towards AI-driven life science research. Its data structure allows it to serve as a foundation for other applications, such as predicting novel uses for existing drugs or cell lines, and is a component of the product. As the AI field grows, it is likely that Bioz will notice the community as a workflow tool, especially for automated high-throughput environments. The company seeks to continue its growth and its combination of partnerships with large cloud providers and mainstream vendors.

Aits and the company did not disclose plans for the future, but it aims to advance its vision of making every biomedical discovery faster and more reproducible. In doing so, it is listening to the OpenAI ecosystem as well.